The U.S. Preventive Health Services Task Force Report of 1989 urges primary care physicians to screen patients for alcohol abuse but not for drug abuse, since a validated, brief, palatable screening test for drug abuse does not exist. The goals of this two-year, four-phase study are to produce and validate a brief questionnaire, consisting of no more than five items, that screen conjointly for alcohol and drug abuse among primary care patients of diverse gender and race. In Phase 1 focus groups of professionals, previous substance abusers, and current substance abusers will suggest and critique potential questionnaire items. In Phase 2, the discriminant validity of up to twenty questionnaire items will be assessed initially on 300 known substance abusers in treatment and 300 nonabusers in primary care practices. In Phase 3, the discriminant validity of the most discriminating items from Phase 2 will be assessed for 300 substance abusing and 600 non-abusing primary care patients (Phase 3). For Phase 2 and 3, the Compositite International Diagnostic Instrument-Substance Abuse Modules (CIDI-SAM) will be used as the chief criterion measure of substance abuse. Agreement by all primary care subjects to undergo urine drug screens after questionnaire administration, as determined by 1-in-4 random draw, will enhance the accuracy of self-report. Through both phases, ongoing analysis will allow prompt replacement of poorly predictive items with items modified from other published questionnaires or suggested by the focus groups. Contingency table analysis, logistic regression, and logistic discrimination will be used to guide the selection of items and formulate recommendations on questionnaire administration and scoring. Demographic variables, as well as responses to screening items and interactions, will be assessed as predictor by new statistical methods. The 13-item Marlowe-Crowne scale will control for social desirability. A cross-validation of items between Phase 2 and 3, a split sample analysis in Phase 3, and measures of internal consistency and test-retest reliability (Phase 4) will be performed. To inform appropriate clinical use of the questionnaire, sensitivity, specificity, and positive and negative predictive values will be reported, and optimal criterion values will be established for varying prevalence rates of alcohol and drug abuse. Possible selection bias will be assessed by brief query of non- participants. Analyses will be repeated to assess the effects of evolving diagnostic classification systems and subjects' responses which, according to separate self-report, may be inaccurate. Conjoint questionnaires will produce the first substance abuse screening tool for adults in primary care settings and lay the groundwork for studies on brief interventions for drug abuse in primary care settings.